The Green Fluorescent Protein (GFP) has presented great opportunities for the use of fluorescence for the study of dynamic cellular processes in living cells. It has also presented great challenges for imaging technologies. This Bioengineering Research Grant proposes to develop a high resolution, high-speed fluorescence microscope for imaging single molecules or small aggregates of molecules moving at the speeds of molecular motors. This instrument will provide 3-D movies with the high temporal and spatial resolution that is needed for live cell imaging. It will be based on a fast, sensitive, low noise 640 x 480 CCD camera, wide field illumination and fast focus change. Spatial resolution of 100 nanometers (transverse) will be obtained by the use of computational methods called image restoration or deconvoluion. It will obtain images at rates of 94 frames per second for full frames images and faster for smaller images (188 frames/second) with 65 percent quantum efficiency and 8.5 electron readout noise. It will obtain 3-D images at rates up to 36 high-resolution 3-D images per second. High pixel rates are needed to provide an image at the speeds needed to follow motion at speeds up to 1 to 2 microns per second. This camera operates at about 30 Megapixels/sec, but since the whole sample is illuminated for the full frame time, fluorescence saturation does not limit signals. Signal levels in scanning confocal microscopes are severely limited by fluorescence saturation at these scanning rates. Sensitivity will be high enough to detect and image single molecules in a cultured cell's cytoplasm and to detect and image small clusters of proteins in parts of the cell with higher auto-fluorescence. Fluorescence will be calibrated so that small numbers of fluorescent proteins can be accurately counted. Imaging protocols will be developed and tested for optimizing the tasks of detecting, imaging and counting small numbers of proteins whether stationary or in motion at the speeds of molecular motors. This system will produce one gigabyte of data in less than 20 seconds. Software will be developed for proving nearly instantaneous feedback to the experimenter on the quality of this large stream of data. Sophisticated image restoration, volume visualization and data analysis software will be provided at the microscope for this purpose.